The Agent Skills Directory

[COMMAND_EXECUTION]: The recalc.py script executes system-level commands to perform headless formula recalculation in Excel files.
It uses subprocess.run to call soffice (LibreOffice) with specific arguments to trigger a recalculation macro.
It optionally utilizes the timeout or gtimeout utilities to manage execution duration.
[DYNAMIC_EXECUTION]: The script recalc.py dynamically generates a LibreOffice Basic macro (Module1.xba) at runtime.
The macro is written to the user's local LibreOffice configuration directory and contains code to calculate all formulas, save, and close the workbook.
This generated code is subsequently executed via a command-line call to the office suite.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external spreadsheet files (.xlsx, .csv, etc.), creating a potential surface for indirect prompt injection.
Ingestion points: The agent reads files using pandas.read_excel (as instructed in SKILL.md) and the recalc.py script loads workbooks via openpyxl.load_workbook to scan cell values.
Boundary markers: The instructions do not specify any delimiters or safety warnings for handling untrusted content within spreadsheets.
Capability inventory: The skill has the ability to write files and execute system commands (soffice, timeout) through the recalc.py utility.
Sanitization: There is no evidence of sanitization or filtering of Excel cell content before it is processed by the agent or script.
[PERSISTENCE]: The recalc.py script modifies the local environment by persisting a macro in the LibreOffice application configuration directory (~/.config/libreoffice/ or ~/Library/Application Support/LibreOffice/). While this is a functional requirement for the skill, the macro remains in the user's application configuration across different sessions.

xlsx